Self-Organizing Ternary Intelligent Manufacturing Control Method and System Based on Hypercycle Network

Abstract

The present invention relates to a self-organizing ternary intelligent manufacturing control method and system based on a hypercycle network. The method includes: S100, the device layer network optimizes device-level process parameters during the production and manufacturing task through self-replicating cyclic iteration; S200, the production line / shop floor layer network optimizes production line / shop floor-level process parameters during the production and manufacturing task through mutual catalysis cyclic iteration; S300, the factory layer network optimizes factory-level process parameters during the production and manufacturing task through high-level mutual catalysis cyclic iteration; S400, the device layer network, the production line / shop floor layer network, and the factory layer network are sequentially arranged from bottom to top, and through the horizontal collaboration and vertical iteration optimization of the device layer network, the production line / shop floor layer network, and the factory layer network, self-planning, self-scheduling, and self-control during the production and manufacturing task are realized. The present invention can adapt to large-scale personalized production and manufacturing tasks, quickly respond to and adjust according to the dynamic changes of large batches of orders, and is more intelligent.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent manufacturing, and in particular to a self-organizing ternary intelligent manufacturing control method and system based on a hypercycle network. Background Art

[0002] The globalization that started at the end of the 20th century has driven the transformation of manufacturing from mass customization to mass personalization. In the context of mass personalization, the characteristics of small batches and large batches of orders pose new challenges to the manufacturing industry, and also enable companies with the ability to handle mass personalization orders to occupy an advantageous position in the market competition.

[0003] The concept of a self-organizing intelligent factory originated from the self-organizing manufacturing system (SOMS) proposed in 1994, which mainly aimed at the production planning and scheduling problems widely faced by manufacturing factories at that time. In the subsequent development, the research on the functions of the SOMS system mainly focused on production planning and scheduling and path planning problems, such as production planning and scheduling based on heuristic methods, dynamic scheduling systems for handling emergency order insertions, path planning methods for workshop logistics scheduling, etc. In addition, in recent years, methods based on multi-agent negotiation and multi-agent have also been applied to SOMS, aiming to make the factory have stronger adaptability and robustness. However, these methods do not touch on the core dilemmas faced by self-organizing intelligent factories, such as the modeling of self-organizing architectures and control architectures, the information coordination and communication mechanism between individuals, and heterogeneous group adaptive control methods.

[0004] Existing intelligent / digital factory technologies mainly rely on information systems such as ERP and MES, which are equivalent to planning / deciding the production tasks to be executed by each workshop, production line, and equipment in the factory from top to bottom starting from the order. Combining methods such as robot motion controllers, PLC logic control, and host computer control, feedback control is carried out at each process and during processing. Such a control method actually still treats each device, production line, and workshop as discrete points, resulting in a large amount of redundant data being generated and collected during the production process. Individual devices do not have independent intelligence, and there is a lack of coordination between the front and back processes in the entire production line or workshop. The decision-making of the entire factory is centered on human intelligence, and it is difficult to quickly respond to the dynamically changing demands of mass personalization orders. Summary of the Invention

[0005] In order to solve the above technical problems existing in the prior art, the purpose of the present invention is to provide a self-organizing ternary intelligent manufacturing control method and system based on a hypercycle network, which can adapt to mass personalization production and manufacturing tasks, quickly respond and adjust according to the dynamic changes of large batch orders, and be more intelligent.

[0006] To achieve the above-mentioned invention object, the technical solution of the present invention is as follows:

[0007] In a first aspect, the present invention provides a self-organizing ternary intelligent manufacturing control method based on a hypercycle network, including:

[0008] S100. The device layer network optimizes the device-level process parameters during the production and manufacturing tasks through self-replicating cyclic iteration;

[0009] S200. The production line / workshop layer network optimizes the production line / workshop-level process parameters during the production and manufacturing tasks through mutual catalysis cyclic iteration;

[0010] S300. The factory layer network optimizes the factory-level process parameters during the production and manufacturing tasks through high-level mutual catalysis cyclic iteration;

[0011] S400. The device layer network, the production line / workshop layer network, and the factory layer network are sequentially arranged from bottom to top, and through the horizontal collaboration and vertical iteration optimization of the device layer network, the production line / workshop layer network, and the factory layer network, self-planning, self-scheduling, and self-control during the production and manufacturing tasks are realized.

[0012] According to one aspect of the present invention, the S100 includes:

[0013] S101. The execution module executes a production and manufacturing task once according to a preset process and actions;

[0014] S102. Record the original production and processing manufacturing data T generated during the task process;

[0015] S103. Extract the key feature vectors of the original production and processing manufacturing data T, and the control module converts the key feature vectors of the original production and processing manufacturing data T into self-replicating catalytic information I;

[0016] S104. The catalytic information I guides the execution module to execute the next production and manufacturing task, and the S102-S104 are repeatedly iterated for continuous feedback optimization until the best device-level process parameters are formed.

[0017] According to one aspect of the present invention, the S200 includes:

[0018] S201. The device Task1 performs a self-replicating cycle according to the S100;

[0019] S202. Record the original production and processing manufacturing data T1 generated during the self-replicating cycle of the device Task1;

[0020] S203. Extract the key feature vectors of the original production and processing manufacturing data T1 and convert them into catalytic information I1;

[0021] S204. The device Task2 adjusts its own self-replication cycle process according to the catalytic information I1;

[0022] S205. Record the original production and manufacturing data T2 generated during the self-replication cycle process of the device Task2;

[0023] S206. Extract the key feature vectors of the original production and manufacturing data T2 and convert them into catalytic information I2;

[0024] S207. The device Task1 adjusts its own self-replication cycle process according to the catalytic information I2;

[0025] S208. Repeatedly iterate S202 - S207 for continuous feedback optimization. The device Task1 and the device Task2 promote each other until the best production line / shop floor-level process parameters are formed.

[0026] According to one aspect of the present invention, the S300 includes:

[0027] S301. The production line / shop floor Task3 performs a self-replication cycle according to S200;

[0028] S302. Record the original production and manufacturing data T3 generated during the self-replication cycle process of the production line / shop floor Task3;

[0029] S303. Extract the key feature vectors of the original production and manufacturing data T3 and convert them into catalytic information I3;

[0030] S304. The production line / shop floor Task4 adjusts its own self-replication cycle process according to the catalytic information I3;

[0031] S305. Record the original production and manufacturing data T4 generated during the self-replication cycle process of the production line / shop floor Task4;

[0032] S306. Extract the key feature vectors of the original production and manufacturing data T4 and convert them into catalytic information I4;

[0033] S307. The production line / shop floor Task3 adjusts its own self-replication cycle process according to the catalytic information I4;

[0034] S308. Repeatedly iterate S302 - S307 for continuous feedback optimization. The production line / shop floor Task3 and the production line / shop floor Task4 promote each other until the best factory-level process parameters are formed.

[0035] According to one aspect of the present invention, extracting the key feature vectors of the original production and manufacturing data and converting them into catalytic information includes:

[0036] Extracting the features of the key parameters of the original production and manufacturing data and forming key feature vectors;

[0037] Processing and transforming the key feature vectors through feedforward and feedback methods to generate catalytic information.

[0038] According to one aspect of the present invention, extracting the features of the key parameters of the original production and manufacturing data includes: predetermining in advance, through expert knowledge, the main parameters that affect the performance of equipment during the production and manufacturing tasks.

[0039] According to one aspect of the present invention, the feedforward and feedback methods are implemented based on traditional cybernetics.

[0040] According to one aspect of the present invention, in S400 during the production and manufacturing tasks, the factory layer network, the production line / workshop layer network, and the equipment layer network successively and autonomously plan their respective corresponding production and manufacturing tasks, autonomously schedule their respective corresponding resources, and autonomously control their production and manufacturing behaviors and processes from top to bottom.

[0041] In a second aspect, the present invention provides a system for implementing the self-organizing ternary intelligent manufacturing control method based on a hypercycle network as described above, including: an equipment layer network, a production line / workshop layer network, and a factory layer network arranged successively from bottom to top,

[0042] Multiple of the equipment layer networks serve as nodes of the production line / workshop layer network and are connected in sequence to form the production line / workshop layer network;

[0043] Multiple of the production line / workshop layer networks serve as nodes of the factory layer network and are connected in sequence to form the factory layer network;

[0044] The equipment layer network, the production line / workshop layer network, and the factory layer network achieve self-planning, self-scheduling, and self-control during the production and manufacturing tasks through their respective cyclic structures.

[0045] According to another aspect of the present invention, the structures of the equipment layer network, the production line / workshop layer network, and the factory layer network are respectively a self-replicating cyclic network, a mutual catalytic cyclic network, and a high-level mutual catalytic cyclic network in sequence.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] According to the solution of the present invention, the traditional manufacturing system is transformed from a passive executor into an intelligent manufacturing control system with three - layer horizontal collaboration and vertical iterative optimization. The system automatically responds to the demands of external small - batch and large - batch orders and the real - time changes of resources, and automatically completes real - time response internally, becoming a flexible production network that can co - evolve.

[0048] According to a solution of the present invention, a self - similar network system constructed by intelligent unit modules formed by intelligent node units and three - layer interfaces provides the possibility for the standardized construction of personalized production manufacturing systems. Under this system, all processing units (equipment), production lines, and factories have the ability to interconnect and form a complete production network, minimizing the personalized customization part and software development required for establishing the production system. This method re - divides the production process into "cells, tissues, and organs" required for self - similar production, providing design and development standards for networked large - scale production. The production line constructed according to this standard will form a stable - evolving system relying on a three - layer hypercycle network and have the ability to construct production systems of various scales. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0050] Figure 1 Schematically showing the network structure diagram of the hypercycle theory provided by the embodiment of the present invention;

[0051] Figure 2 Schematically showing the structure diagram of a self - organizing ternary intelligent manufacturing control system based on a hypercycle network provided by the embodiment of the present invention;

[0052] Figure 3 Schematically showing the self - replication cycle network of the device layer provided by the embodiment of the present invention;

[0053] Figure 4 Schematically showing the mutual - catalysis cycle network of the production line / workshop layer provided by the embodiment of the present invention;

[0054] Figure 5 Schematically showing the high - level mutual - catalysis cycle network of the factory layer provided by the embodiment of the present invention;

[0055] Figure 6 Schematically showing the implementation flowchart of a self - organizing ternary intelligent manufacturing control method based on a hypercycle network provided by the embodiment of the present invention;

[0056] Figure 7 The flowchart of step S200 provided by the embodiment of the present invention is schematically shown. DETAILED DESCRIPTION

[0057] The description of the embodiments in this specification should be combined with the corresponding drawings, which should be considered a complete part of this specification. In the drawings, the shapes and thicknesses of the embodiments may be exaggerated and indicated for simplicity or convenience. Furthermore, the various structural components in the drawings will be described separately. It is worth noting that components not shown in the drawings or not described in words are known to those of ordinary skill in the art.

[0058] The description of the embodiments herein and any references to directions and orientations are for ease of description only and are not to be construed as limiting the scope of the present invention. The following description of the preferred embodiments may involve combinations of features, which may exist independently or in combination. The present invention is not specifically limited to the preferred embodiments. The scope of the present invention is defined by the claims.

[0059] This invention uses the hypercycle theory from the "Origin of Life Theory" as its core and applies it to the network construction of intelligent manufacturing control systems (intelligent factory control systems), thereby obtaining an intelligent manufacturing control system with a self-organizing mechanism. For ease of understanding, the following is a brief introduction to the original hypercycle theory.

[0060] To explain the "chicken and egg" paradox of the origin of life (the precise production of nucleic acids requires highly specialized enzymes, while highly specialized enzymes require precisely defined nucleic acids to produce them, thus creating the "chicken and egg" paradox), starting in the 1970s, a series of theories centered on the origin of life, such as the RNA World theory, have emerged. These theories, such as autopoiesis, the chemoton, and the autocatalytic set, are not widely accepted as such. The most comprehensive of these theories is the hypercycle theory. The original hypercycle core model suffers from several major issues: 1) The loop structure lacks robustness. If a node in the loop malfunctions or errors, damage to the entire loop or the resulting errors accumulate along the loop. 2) The core structure is considered "too artificial," with only limited biochemical evidence and examples.

[0061] The meaning of hyperloop is "loop within loop", and its essence is the structural form of self-organizing system, which is mainly composed of self-replication at the individual level and information iteration loop at the overall level. Figure 1 As shown,Figure 1 In this, I represents the individual self-replication process, and E represents the catalyst. Overall, I and E constitute a catalytic cycle. Figure 1 Only two layers of cycles are drawn in this figure. In an actual self-organizing system (the original research object is a biological macromolecule system), the two-layer cycle structure can further form a three-layer cycle structure.

[0062] As Figure 2 As shown, a self-organizing ternary intelligent manufacturing control system based on a hypercycle network disclosed in this embodiment is used to implement the self-organizing ternary intelligent manufacturing control method of this embodiment. The overall topological structure of this self-organizing ternary intelligent manufacturing control system is a three-layer network. Specifically, this three-layer network structure is as follows: a device layer network, a production line / workshop layer network, and a factory layer network are sequentially arranged from bottom to top. Among them, multiple device layer networks serve as nodes of the production line / workshop layer network and are sequentially connected to form the production line / workshop layer network. Multiple production line / workshop layer networks serve as nodes of the factory layer network and are sequentially connected to form the factory layer network. The device layer network, the production line / workshop layer network, and the factory layer network achieve self-planning, self-scheduling, and self-control during the production and manufacturing task process through their respective cycle structures. Correspondingly, the structures of the device layer network, the production line / workshop layer network, and the factory layer network are respectively a self-replication cycle network, a mutual-catalysis cycle network, and a high-level mutual-catalysis cycle network.

[0063] It can be seen from this that there are self-similar characteristics between the networks at the above-mentioned various levels. On the one hand, the above three-layer network has a fractal structure in terms of structure, that is, the lower-layer network is the node that composes its upper-layer network, and the nodes and connections of each layer of the network are similar. On the other hand, the above three-layer network has the functions of self-planning, self-scheduling, and self-control in terms of function. The device layer network, the production line / workshop layer network, and the factory layer network as a whole constitute a cycle structure. Each cycle represents the process of completing a product production and manufacturing task, that is, the processing and production manufacturing process from raw material S to product P (S→P).

[0064] As Figure 2-3As shown, for the self-replicating network at the equipment layer, during the production of products by the machine, raw data is generated. After extracting the key feature vectors, self-replicating catalytic information I can be generated. During this production process, the previous information and the current information are combined into the current catalytic information I. In the current production, I first combines with the raw material S to form the intermediate product IS, then the intermediate product IP is generated during the processing, and finally, this catalytic information guides the next cycle in the form of feedback and outputs the product P. By continuously repeating the above self-replicating cycle, the catalytic information accumulates and is timely feedback-regulated, completing the optimization of the equipment production process under the guidance of the catalytic information. Additionally, the optimization here refers to the description from the performance perspective. From the functional perspective, the self-replicating network at the equipment layer has the capabilities of autonomous planning, scheduling, and control. Among them, the control mainly refers to the motion control of the machine equipment. The planning is related to the upper level, and under the overall production task allocation, it autonomously plans its own production tasks and autonomously schedules the required resources.

[0065] As Figure 2 and Figure 4 shown, for the mutual catalytic network at the production line / workshop layer, at the production line / workshop level, the "self-replicating networks" of each process form the nodes in the "mutual catalytic network", that is, the equipment production process is regarded as an individual in the production line / workshop layer network. The material flow direction constitutes a feedforward structure (solid line) with a causal relationship, and the causal and correlation relationships such as quality, efficiency, and control constitute the feedback link closed-loop (dashed line) in the "mutual catalytic network", jointly forming the mutual catalytic network at the production line / workshop layer. Similar to the self-replicating network at the equipment layer, that is, during the production process of the production line, raw data is generated. After extracting the key feature vectors, self-replicating catalytic information I can be generated. During the current production process, the previous information and the current information are combined into the current catalytic information I. In the current production, I first combines with the raw material S to form the intermediate product IS, then the intermediate product IP is generated during the processing, and finally, this catalytic information guides the next cycle in the form of feedback and outputs the product P. The catalytic information generated at the production line level is represented by Ipi (IMi is used at the equipment level). Under the guidance and scheduling of Ipi, the planning function of the self-replicating network at the equipment level is completed; at the same time, the production line level network itself also has the functions of autonomous planning, scheduling, and control, and its specific implementation process is similar to that at the equipment level.

[0066] As Figure 2 and Figure 5As shown in the figure, for the high-level mutual catalysis network at the factory level, at the factory level, the "mutual catalysis network" of each production line forms a node in the "high-level mutual catalysis network," that is, the production process of the production line is regarded as an individual in the factory-level network. The order information flow of each factory, the material supply flow in the supply chain, and the quality flow of upstream and downstream product quality form a feedforward structure with causal relationships (solid line). The causal and correlation relationships such as quality, efficiency, and control form the feedback link closed loop (dashed line) in the "mutual catalysis network," together forming the factory-level mutual catalysis network. Its specific implementation process is similar to that of the production line-level mutual catalysis network.

[0067] At this point, the controllers at each level in the ternary intelligent manufacturing control system based on the hyperloop network have formed an overall three-layer network structure. Each node in each layer of the network has a certain degree of autonomous intelligence. At the same time, the links in the network are composed of key information generated in the production process of each level (such as process parameter indicators, etc.), so that the controller network as a whole has the functions of self-planning, self-scheduling, and self-control.

[0068] In the technical solutions of the above embodiments, by establishing an intelligent manufacturing control system based on a hyperloop network, each level is connected through the hyperloop network, improving the efficiency of information transmission and replacing the method of manual decision-making, thereby meeting the response speed of large-scale personalized production and manufacturing tasks. Hyperloop theory is applied to the network construction of intelligent manufacturing control systems. A three-layer self-similar network formed by intelligent units and interfaces is connected vertically and horizontally, improving the collaborative efficiency of workshops, factories, and industrial chains, and avoiding the waste of resources in each link.

[0069] like Figure 6 As shown, this embodiment also discloses a method for implementing the self-organizing ternary intelligent manufacturing control system based on the hyperloop network. The process of implementing the self-organizing ternary intelligent manufacturing control method based on the hyperloop network specifically includes the following steps:

[0070] S100, the device layer network optimizes the device-level process parameters in the production and manufacturing process through self-replication loop iteration;

[0071] S200, the production line / workshop layer network optimizes the production line / workshop level process parameters during the production and manufacturing tasks through mutual catalytic cycles and iterations;

[0072] S300, the factory-level network optimizes factory-level process parameters during the production and manufacturing process through high-level mutual catalysis cycles and iterations;

[0073] The S400, device layer network, production line / workshop layer network, and factory layer network are arranged in sequence from bottom to top. Through the horizontal collaboration and vertical iterative optimization of the device layer network, production line / workshop layer network, and factory layer network, self-planning, self-scheduling, and self-control are realized during the production manufacturing task process.

[0074] Specifically, in step S400, during the production manufacturing task process, the factory layer network, production line / workshop layer network, and device layer network successively and independently plan their respective production manufacturing tasks, independently schedule their respective resources, and independently control their respective production manufacturing behaviors and processes from top to bottom.

[0075] Through the above solution, the order tasks of product production and manufacturing successively pass through the factory layer network, production line / workshop layer network, and device layer network from top to bottom. Through the internal self-loop of the device layer network, the device-level process parameters during the execution of the product production task are continuously optimized. Through the mutual catalytic loop within the factory layer network and production line / workshop layer network, that is, the relationship of mutual cooperation and mutual influence between different production lines and different devices during the product production task process is continuously optimized, promoted, and balanced. Finally, the optimal factory-level, production line-level, and device-level process parameters are obtained, so as to realize the self-planning of the product production and processing tasks corresponding to each level from top to bottom, and independently schedule the required resources such as manpower, machinery and equipment, raw materials, production and processing time, etc. according to their respective tasks, use the controllers of each level network to control and monitor the specific execution actions of production manufacturing, and adjust the specific production process according to the corresponding task objectives.

[0076] According to an embodiment of the present invention, the specific implementation process of the device layer network optimizing the device-level process parameters during the production manufacturing task process through self-replicating cyclic iteration includes the following steps:

[0077] S101. An execution module, such as a processing machine, an intelligent robot, etc., executes a production manufacturing task once according to a preset process and actions;

[0078] S102. Record the original production and processing manufacturing data T generated during the task; the original production and processing manufacturing data T here includes various production process parameters such as equipment performance parameter indicators and processing parameters involved in the process of the processing machine and equipment executing the production and processing manufacturing task.

[0079] S103. Extract the key feature vectors of the original production and processing manufacturing data T, and the control module converts the key feature vectors of the original production and processing manufacturing data T into self-replicating catalytic information I;

[0080] S104. The catalytic information I guides the execution module to execute the next production and manufacturing task, and steps S102 - S104 are iteratively repeated for continuous feedback optimization until the best equipment-level process parameters are formed.

[0081] According to an embodiment of the present invention, as Figure 7 shown, the specific implementation process of the production line / workshop-level network in step S200 for optimizing the production line / workshop-level process parameters during the production and manufacturing task through mutual catalytic cyclic iteration includes the following steps:

[0082] S201. The equipment Task1 performs a self-replication cycle according to the specific implementation process of step S100.

[0083] S202. Record the original production and manufacturing data T1 generated during the self-replication cycle of the equipment Task1; the original production and manufacturing data T1 here includes various production process parameters such as equipment performance parameter indicators and processing parameters involved in the processing machine equipment Task1 during the execution of the production and manufacturing task.

[0084] S203. Extract the key feature vectors of the original production and manufacturing data T1 and convert them into catalytic information I1.

[0085] S204. The equipment Task2 adjusts its own self-replication cycle according to the catalytic information I1.

[0086] S205. Record the original production and manufacturing data T2 generated during the self-replication cycle of the equipment Task2; the original production and manufacturing data T2 here includes various production process parameters such as equipment performance parameter indicators and processing parameters involved in the processing machine equipment Task2 during the execution of the production and manufacturing task.

[0087] S206. Extract the key feature vectors of the original production and manufacturing data T2 and convert them into catalytic information I2.

[0088] S207. The equipment Task1 adjusts its own self-replication cycle according to the catalytic information I2.

[0089] S208. Repeat steps S202 - S207 for continuous feedback optimization. The equipment Task1 and the equipment Task2 promote each other until the best production line / workshop-level process parameters are formed. That is to say, through the above mutual catalytic process, the horizontal coordination and cooperation relationship between the equipment Task1 and the equipment Task2 during the production task can be realized, and the task objectives at the production line / workshop level can be better achieved.

[0090] According to an embodiment of the present invention, the specific implementation process of the factory-level network optimizing the factory-level process parameters during the production and manufacturing tasks through high-level mutual catalytic cycle iteration in step S300 includes the following steps:

[0091] S301. The production line / workshop Task3 performs a self-replication cycle according to the specific implementation process of step S200;

[0092] S302. Record the original production and manufacturing data T3 generated during the self-replication cycle of the production line / workshop Task3; the original production and manufacturing data T3 here includes various performance parameter indicators, processing parameters, etc. of the production process parameters involved in the collaborative operation between different devices involved in the production line Task3 during the execution of the production and manufacturing tasks.

[0093] S303. Extract the key feature vectors of the original production and manufacturing data T3 and convert them into catalytic information I3;

[0094] S304. The production line / workshop Task4 adjusts its own self-replication cycle according to the catalytic information I3;

[0095] S305. Record the original production and manufacturing data T4 generated during the self-replication cycle of the production line / workshop Task4; the original production and manufacturing data T4 here includes various performance parameter indicators, processing parameters, etc. of the production process parameters involved in the collaborative operation between different devices involved in the production line Task4 during the execution of the production and manufacturing tasks.

[0096] S306. Extract the key feature vectors of the original production and manufacturing data T4 and convert them into catalytic information I4;

[0097] S307. The production line / workshop Task3 adjusts its own self-replication cycle according to the catalytic information I4;

[0098] S308. Repeat the iterative steps S302 to S307 for continuous feedback optimization. The production line / workshop Task3 and the production line / workshop Task4 promote each other until the best factory-level process parameters are formed. That is to say, through the above mutual catalytic process, the horizontal coordination and cooperation relationship between the production line Task3 and the production line Task4 during the production task can be realized, and the task goals at the factory level can be better achieved.

[0099] Specifically, the specific implementation process of extracting the key feature vectors of the original production and manufacturing data in the above steps S100, S200, and S300 and converting them into catalytic information includes: extracting the features of the key parameters of the original production and manufacturing data and forming key feature vectors; processing and transforming the key feature vectors through feedforward and feedback methods to generate catalytic information. In step S100, the conversion process of catalytic information is mainly a process of converting the feature vector of the current production task of the equipment into information useful for the next production task (optimized processing process parameters). In steps S200 and S300, the conversion process of catalytic information is mainly a process of converting the feature vector of processing operation or task A into information useful for another processing operation or task B (optimized processing process parameters).

[0100] Among them, the specific implementation process of extracting the features of the key parameters of the original production and manufacturing data includes: determining in advance the main parameters that affect the performance of the equipment during the production and manufacturing tasks through expert knowledge. In some other embodiments, statistical methods such as principal component analysis PCA (unsupervised dimensionality reduction method), LDA, or SVD can also be used for key feature extraction, which is applicable to data with different probability distributions. Among them, principal component analysis PCA includes the following steps: 1) After organizing the data, calculate the covariance matrix; 2) Calculate the eigenvalues and eigenvectors of the covariance matrix; 3) Sort the eigenvectors according to the eigenvalues; 4) Select the main components according to the sorting results and the number of principal components required. In other embodiments, a neural network-based method can also be used for key feature extraction. For example, for a two-dimensional (three-dimensional) data structure such as an image, through the "small window" of the convolutional kernel, each two-dimensional plane data in the image is scanned and aggregated row by row, and feature extraction is performed layer by layer; each hidden layer in the neural network is a process of extracting picture information once. As the network deepens, higher-level and lower-dimensional information is gradually extracted; finally, a feature vector that meets the dimensionality requirements is generated. Through the feature extraction of these methods, the data can be reduced in dimension, the calculation amount in the ternary intelligent manufacturing network can be reduced, and redundant data can be excluded.

[0101] The feedforward and feedback methods are based on traditional cybernetics, and a feasible specific implementation is the Kalman filter.

[0102] Exemplarily, taking the closed-loop structure in the fixed-weight and fixed-length system in the metallurgical industry as an example, the specific implementation processes of the above steps S100 and S200 will be further described. During the self-replicating cycle, the execution module consists of a fixed-length cutting system and a weighing device, and the control module consists of a fixed-weight system. The specific process is as follows: a1) The fixed-length cutting system in the execution module first performs a cutting operation, and the weighing device performs a weight measurement; a2) During the process of performing the cutting and weighing tasks, data on the length L1 and weight W1 of the billet are generated and integrated into the catalytic information vector T(L1, W1); a3) The fixed-weight system adjusts the weight estimate W2 and the length adjustment amount △L2 of the next billet according to T(L1, W1), and integrates them into the catalytic information enzyme vector I(△L2, W2); a4) The catalytic information enzyme vector I(△L2, W2) guides the execution module to perform the next billet cutting and weighing operation, and the process of a2) -> a4) is repeated. Through continuous positive feedback in the self-replicating closed-loop cycle, the billet weight is stabilized near the standard value until the goal of cost reduction and quality improvement is achieved.

[0103] During the process of the mutual catalytic cycle, equipment Task1 refers to the operation of processing molten steel into billets by a continuous casting machine, and its evolutionary goal is to make the density of the billets uniform; equipment Task2 refers to the operation of fixed-weight cutting by a fixed-weight and fixed-length system, and its evolutionary goal is to make the weight of each billet uniform. The specific process is as follows: b1) The main body of the continuous casting machine performs the operation of processing molten steel into billets; b2) When performing the billet processing task, collect core process parameters, such as the mold drawing speed V, the ladle surface height H, the billet temperature T in the secondary cooling zone, and the mold life E; b3) Convert the core process parameters into a catalytic information vector I1 = (V, T, H, E); b4) The fixed-weight and fixed-length system constructs a feedforward model W1 = f(I1 = (V, T, H, E)) based on I1, and optimizes the weight prediction value of the fixed-weight and fixed-length system through the feedforward prediction method; b5) When performing the billet length adjustment amount and weight estimation, generate the estimated values △L and W of the length adjustment amount and weight; b6) The fixed-weight system integrates the billet length adjustment amount △L (variation range) and the billet weight W into catalytic information I2 = (△L, W); b7) The main body of the continuous casting machine adjusts the above core process parameters during the billet processing according to I2; b8) Repeat the process of b2) -> b7) for feedback optimization until the overall density of the billets is uniform and the weight of each billet is stable, so as to achieve the goal of cost reduction, quality improvement and efficiency increase in the overall continuous casting process. Among them, the feedforward prediction method is specifically: according to the mapping relationship between the historical data (V, T, H, E) of the core process parameters and the actual billet weight W, establish a feedforward model W1 = f(I1 = (V, T, H, E)) through the regression method, and output the billet weight prediction value W1 = f(V, T, H, E); under the assumption that the billet density is uniform and the shape is uniform, calculate the theoretical value W2 = ρsl of the billet weight. The process of feedback optimization adjustment is specifically: the actual measured billet weight after performing the cutting operation is W3, different weights are assigned to W1, W2, and W3 respectively, and the predicted value W of the billet weight when performing the next cutting operation without adjusting the billet length is calculated (that is, the one-dimensional form of the Kalman filter); then according to the difference between W and the target weight W', divide it by the product of the density ρ and the cross-sectional area s, and convert it into the length adjustment amount ΔL' of the next cutting task.

[0104] The sequence numbers of the above-mentioned various steps involved in the method of the present invention do not mean the sequence of method execution. The execution sequence of each step should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiment of the present invention.

[0105] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A self-organizing ternary intelligent manufacturing control method based on a hypercycle network, comprising: S100. The device layer network optimizes the device-level process parameters during the production manufacturing task through self-replicating cyclic iteration, including: S101. The execution module executes a production manufacturing task once according to a preset process and actions; S102. Record the original production processing and manufacturing data T generated during the task; S103. Extract the key feature vectors of the original production processing and manufacturing data T, and the control module converts the key feature vectors of the original production processing and manufacturing data T into self-replicating catalytic information I; S104. The self-replicating catalytic information I guides the execution module to execute the next production manufacturing task, and repeat and iterate S102 - S103 for continuous feedback optimization until the best device-level process parameters are formed; S200. The production line / workshop layer network optimizes the production line / workshop-level process parameters during the production manufacturing task through mutual catalytic cyclic iteration; S300. The factory layer network optimizes the factory-level process parameters during the production manufacturing task through high-level mutual catalytic cyclic iteration; S400. The device layer network, the production line / workshop layer network, and the factory layer network are set up in sequence from bottom to top, and through the horizontal coordination and vertical iterative optimization of the device layer network, the production line / workshop layer network, and the factory layer network, self-planning, self-scheduling, and self-control during the production manufacturing task process are realized.

2. The method according to claim 1, wherein The S200 includes: S201. Device Task1 performs a self-replicating cycle according to S100; S202. Record the original production processing and manufacturing data T1 generated during the self-replicating cycle of Device Task1; S203. Extract the key feature vectors of the original production processing and manufacturing data T1 and convert them into catalytic information I1; S204. Device Task2 adjusts its own self-replicating cycle process according to the catalytic information I1; S205. Record the original production processing and manufacturing data T2 generated during the self-replicating cycle of Device Task2; S206. Extract the key feature vectors of the original production processing and manufacturing data T2 and convert them into catalytic information I2; S207. Device Task1 adjusts its own self-replicating cycle process according to the catalytic information I2; S208. Repeat and iterate S202 - S207 for continuous feedback optimization, and Device Task1 and Device Task2 promote each other until the best production line / workshop-level process parameters are formed.

3. The method according to claim 1, wherein The S300 includes: S301. The production line / workshop Task3 performs a self-replicating cycle according to S200; S302. Record the original production processing and manufacturing data T3 generated during the self-replicating cycle of the production line / workshop Task3; S303. Extract the key feature vectors of the original production processing and manufacturing data T3 and convert them into catalytic information I3; S304. The production line / workshop Task4 adjusts its own self-replicating cycle process according to the catalytic information I3; S305. Record the original production and manufacturing data T4 generated during the self-replication cycle of the production line / workshop Task4; S306. Extract the key feature vectors of the original production and manufacturing data T4 and convert them into catalytic information I4; S307. The production line / workshop Task3 adjusts its own self-replication cycle process according to the catalytic information I4; S308. Repeat and iterate S302 - S307 for continuous feedback optimization. The production line / workshop Task3 and the production line / workshop Task4 promote each other until the optimal factory-level process parameters are formed.

4. The method according to any one of claims 1 to 3, characterized in that, The extraction of the key feature vectors of the original production and manufacturing data and the conversion into catalytic information include: Extract the features of the key parameters of the original production and manufacturing data and form key feature vectors; Process and transform the key feature vectors through feedforward and feedback methods to generate catalytic information.

5. The method according to claim 4, characterized in that The extraction of the features of the key parameters of the original production and manufacturing data includes: predetermine the main parameters affecting the equipment performance during the production and manufacturing task process through expert knowledge.

6. The method according to claim 4, characterized in that, The feedforward and feedback methods are implemented based on traditional cybernetics.

7. The method according to claim 1, wherein In S400, during the production and manufacturing task process, the factory layer network, the production line / workshop layer network, and the equipment layer network successively and autonomously plan their respective corresponding production and manufacturing tasks, autonomously schedule their respective corresponding resources, and autonomously control their respective production and manufacturing behaviors and processes from top to bottom.

8. A system for implementing the self-organizing three-element intelligent manufacturing control method based on a hypercycle network as described in any one of claims 1-7, characterized in that, Including: An equipment layer network, a production line / workshop layer network, and a factory layer network are sequentially arranged from bottom to top. Multiple of the equipment layer networks serve as nodes of the production line / workshop layer network and are sequentially connected to form the production line / workshop layer network; Multiple of the production line / workshop layer networks serve as nodes of the factory layer network and are sequentially connected to form the factory layer network; The equipment layer network, the production line / workshop layer network, and the factory layer network achieve self-planning, self-scheduling, and self-control during the production and manufacturing task process through their respective cyclic structures.

9. The system according to claim 8, wherein The structures of the equipment layer network, the production line / workshop layer network, and the factory layer network are respectively a self-replication cyclic network, a mutual-catalysis cyclic network, and a high-level mutual-catalysis cyclic network in sequence.

Images